On the effect of spatial variability and support on validation of remote sensing observations of CO2

J.M. Tadić and A.M. Michalak

Satellite observations of atmospheric CO2 are often validated against other CO2 measurements, such as in situ airborne observations, other existing space-borne instruments, or ground-based remote sensing instruments. Here we examine the extent to which such comparisons are useful as a validation strategy for space-based instruments, by exploring the degree to which horizontal variability in atmospheric CO2 and the difference in the spatial footprint of these various types of observations can impact the intercomparability of observations. We use OCO-2, GOSAT, TCCON, and airborne observations as test cases, and find that the differences in the observed signal among measurement platforms resulting from horizontal variability and differences in the spatial support can be substantial. This work is especially informative given the ongoing efforts to refine the OCO-2 retrieval algorithm -- efforts that are informed by intercomparisons with the types of CO2 observations examined here.

Figure: Horizontal projections of the spatial supports of TCCON (green), aircraft (black), GOSAT (red) and OCO-2 satellite (blue) observations for (a) Walnut Grove and (b) Petaluma; vertical simulated CO2 profiles encountered (only a portion of the column shown) and corresponding XCO2 values for full profiles for (c) Walnut Grove and (d) Petaluma, based on one randomly selected conditional simulation for each site; histograms of differences in the observed XCO2 between TCCON (the “true” XCO2 at the centerpoint of the simulated domain) and the other three platforms based on 1000 conditional realizations for (e) Walnut Grove and (f) Petaluma (the histograms of aircraft/TCCON differences (grey) are overlaid with translucent TCCON/GOSAT (red) and TCCON/OCO-2 (blue) histograms) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).


Validation of ground-based and satellite remote sensing CO2 observations involves comparisons among platforms and with in situ airborne measurements. Several factors unrelated to observational errors can lead to mismatches between measurements, and must be assessed to avoid misinterpreting actual differences in observed values as errors. Here we explore the impact of CO2 horizontal variability and differences in the spatial support of measurements. Case studies based on flights over Walnut Grove and Petaluma, California, are used to compare hypothetical airborne, TCCON, GOSAT, and OCO-2 measurements. We find that high CO2 variability can lead to differences in inferred XCO2 (1) of over 0.5 ppm between airborne and remote sensing observations, due to the spatial mismatch between spiral flight trajectories and atmospheric columns, and (2) of up to 0.3 ppm among remote sensing platforms, due to differences in the spatial support of observations. Horizontal CO2 variability must therefore be considered in intercomparisons aimed at validation of remote sensing observations.

Tadić, J.M., A.M. Michalak (2016) "On the effect of spatial variability and support on validation of remote sensing observations of CO2", Atmospheric Environment, 132, 309-316, doi:10.1016/j.atmosenv.2016.03.014.